Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204104158> ?p ?o ?g. }
- W3204104158 endingPage "632" @default.
- W3204104158 startingPage "622" @default.
- W3204104158 abstract "A large labeled dataset is a key to the success of supervised deep learning, but for medical image segmentation, it is highly challenging to obtain sufficient annotated images for model training. In many scenarios, unannotated images are abundant and easy to acquire. Self-supervised learning (SSL) has shown great potentials in exploiting raw data information and representation learning. In this paper, we propose Hierarchical Self-Supervised Learning (HSSL), a new self-supervised framework that boosts medical image segmentation by making good use of unannotated data. Unlike the current literature on task-specific self-supervised pretraining followed by supervised fine-tuning, we utilize SSL to learn task-agnostic knowledge from heterogeneous data for various medical image segmentation tasks. Specifically, we first aggregate a dataset from several medical challenges, then pre-train the network in a self-supervised manner, and finally fine-tune on labeled data. We develop a new loss function by combining contrastive loss and classification loss, and pre-train an encoder-decoder architecture for segmentation tasks. Our extensive experiments show that multi-domain joint pre-training benefits downstream segmentation tasks and outperforms single-domain pre-training significantly. Compared to learning from scratch, our method yields better performance on various tasks (e.g., (+0.69%) to (+18.60%) in Dice with (5%) of annotated data). With limited amounts of training data, our method can substantially bridge the performance gap with respect to denser annotations (e.g., (10%) vs. (100%) annotations)." @default.
- W3204104158 created "2021-10-11" @default.
- W3204104158 creator A5019362565 @default.
- W3204104158 creator A5047868177 @default.
- W3204104158 creator A5059912382 @default.
- W3204104158 creator A5060901632 @default.
- W3204104158 creator A5073124436 @default.
- W3204104158 creator A5075201879 @default.
- W3204104158 creator A5089809484 @default.
- W3204104158 date "2021-01-01" @default.
- W3204104158 modified "2023-10-02" @default.
- W3204104158 title "Hierarchical Self-supervised Learning for Medical Image Segmentation Based on Multi-domain Data Aggregation" @default.
- W3204104158 cites W1901129140 @default.
- W3204104158 cites W2046289434 @default.
- W3204104158 cites W2100296123 @default.
- W3204104158 cites W2194775991 @default.
- W3204104158 cites W2308529009 @default.
- W3204104158 cites W2321533354 @default.
- W3204104158 cites W2625559849 @default.
- W3204104158 cites W2750023899 @default.
- W3204104158 cites W2750925197 @default.
- W3204104158 cites W2804047627 @default.
- W3204104158 cites W2806321514 @default.
- W3204104158 cites W2904204744 @default.
- W3204104158 cites W2963420272 @default.
- W3204104158 cites W2963609467 @default.
- W3204104158 cites W2964744899 @default.
- W3204104158 cites W2974089460 @default.
- W3204104158 cites W2979708377 @default.
- W3204104158 cites W2979888373 @default.
- W3204104158 cites W2979907638 @default.
- W3204104158 cites W2979967613 @default.
- W3204104158 cites W2991391304 @default.
- W3204104158 cites W2998097997 @default.
- W3204104158 cites W3015788359 @default.
- W3204104158 cites W3035524453 @default.
- W3204104158 cites W3093130481 @default.
- W3204104158 cites W3097337894 @default.
- W3204104158 cites W343636949 @default.
- W3204104158 doi "https://doi.org/10.1007/978-3-030-87193-2_59" @default.
- W3204104158 hasPublicationYear "2021" @default.
- W3204104158 type Work @default.
- W3204104158 sameAs 3204104158 @default.
- W3204104158 citedByCount "6" @default.
- W3204104158 countsByYear W32041041582022 @default.
- W3204104158 countsByYear W32041041582023 @default.
- W3204104158 crossrefType "book-chapter" @default.
- W3204104158 hasAuthorship W3204104158A5019362565 @default.
- W3204104158 hasAuthorship W3204104158A5047868177 @default.
- W3204104158 hasAuthorship W3204104158A5059912382 @default.
- W3204104158 hasAuthorship W3204104158A5060901632 @default.
- W3204104158 hasAuthorship W3204104158A5073124436 @default.
- W3204104158 hasAuthorship W3204104158A5075201879 @default.
- W3204104158 hasAuthorship W3204104158A5089809484 @default.
- W3204104158 hasBestOaLocation W32041041582 @default.
- W3204104158 hasConcept C108583219 @default.
- W3204104158 hasConcept C111919701 @default.
- W3204104158 hasConcept C118505674 @default.
- W3204104158 hasConcept C119857082 @default.
- W3204104158 hasConcept C134306372 @default.
- W3204104158 hasConcept C136389625 @default.
- W3204104158 hasConcept C153180895 @default.
- W3204104158 hasConcept C154945302 @default.
- W3204104158 hasConcept C162324750 @default.
- W3204104158 hasConcept C187736073 @default.
- W3204104158 hasConcept C2776145971 @default.
- W3204104158 hasConcept C2780451532 @default.
- W3204104158 hasConcept C33923547 @default.
- W3204104158 hasConcept C36503486 @default.
- W3204104158 hasConcept C41008148 @default.
- W3204104158 hasConcept C50644808 @default.
- W3204104158 hasConcept C58973888 @default.
- W3204104158 hasConcept C59404180 @default.
- W3204104158 hasConcept C89600930 @default.
- W3204104158 hasConceptScore W3204104158C108583219 @default.
- W3204104158 hasConceptScore W3204104158C111919701 @default.
- W3204104158 hasConceptScore W3204104158C118505674 @default.
- W3204104158 hasConceptScore W3204104158C119857082 @default.
- W3204104158 hasConceptScore W3204104158C134306372 @default.
- W3204104158 hasConceptScore W3204104158C136389625 @default.
- W3204104158 hasConceptScore W3204104158C153180895 @default.
- W3204104158 hasConceptScore W3204104158C154945302 @default.
- W3204104158 hasConceptScore W3204104158C162324750 @default.
- W3204104158 hasConceptScore W3204104158C187736073 @default.
- W3204104158 hasConceptScore W3204104158C2776145971 @default.
- W3204104158 hasConceptScore W3204104158C2780451532 @default.
- W3204104158 hasConceptScore W3204104158C33923547 @default.
- W3204104158 hasConceptScore W3204104158C36503486 @default.
- W3204104158 hasConceptScore W3204104158C41008148 @default.
- W3204104158 hasConceptScore W3204104158C50644808 @default.
- W3204104158 hasConceptScore W3204104158C58973888 @default.
- W3204104158 hasConceptScore W3204104158C59404180 @default.
- W3204104158 hasConceptScore W3204104158C89600930 @default.
- W3204104158 hasLocation W32041041581 @default.
- W3204104158 hasLocation W32041041582 @default.
- W3204104158 hasOpenAccess W3204104158 @default.
- W3204104158 hasPrimaryLocation W32041041581 @default.
- W3204104158 hasRelatedWork W1756896031 @default.